Coral Reef Gender & Diversity Analysis
Sam Cheng
01/18/2020
Introduction
The goal of this project is to examine patterns in author diversity (gender/geographic) in coral reef publications over the past 15 years. We selected a range of peer-reviewed scientific journals across different impact factors and levels (incl. Conservation Letters, PNAS, Proceedings of the Royal Society B, Coral Reefs, PlosOne, Marine Policy, Marine Biology, Frontiers in Marine Science, Nature, Science). A team of global coral reef scientists reviewed these articles and extracted relevant data on research area, geography, authorship, and gender.
Notes for co-authors to consider are in bold purple
Summary statistics
This study considered 1683 unique studies published between 2003 and NA. This dataset represents works by 5258 unique authors from 357 different countries
The number of unique authors is going to be wrong as names were not standardized between databases). We need to decide whether we want to spend time cleaning this or not. Variations include Last,First; First Last; Last, First Initial
Figure 1. Growth in coral reef publications over the last 15 years, disaggregated by journal
Author characteristics (gender, diversity)
This study found a maximum of 5258 unique authors in the evidence base. This number represents the maximum number of authors as different publication databases listed authors in different formats and thus we were unable to de-duplicate entries that were not an exact match (e.g. Last Name, First Name vs. Last Name, First Initial). 59 authors could not be located on the internet to confirm gender identity. Authors whose gender identity could not be reliably determined based on the presence of identified pronouns, declarations, or author(s) knowledge - were coded as “undetermined.” The majority of authors in coral reef science are male.
Figure 2. Gender distribution of authors in coral reef science
Figure X. Frequency of male vs. female authors in first, last, and all other author positions
Gender diversity trends
The following displays patterns in gender by looking at 4 metrics: author position (first, last), single author papers, and proportion of publication output contributed by men vs. women) across a number of different variables (publication year, country of affiliation, marine realm studied, study type, times cited, and journal).
Proportion of each gender’s output was calculated by comparing the proportion of identified authorships for each gender in a given article. For example, if a paper had 9 authors, of which 8 were identified to gender, each authorship was granted an equal share of authorship (1/8 each). These fractions were then aggregated at different levels (e.g. countries, study_types, journals) to explore gender contributions. These reflect authorships and not individuals.
Fix figure titles. First graph - proportion of women vs. men as one bar
Trends over time
Figure X. Change in composition of gender in first, last, and single author positions over time
Figure X. Change in proportion of individual contribution of genders over time
Geographic distribution of research
Research was conducted in 94 countries and 18 territories. The most studied countries were Australia (n=458) and United States (n=238). The figure below shows the geographic distribution of research effort in terms of areas under study (using any type of method, e.g. field studies, laboratory, and desk studies).
Figure 2 (alternate). Geographic distribution of study countries with discrete breaks
Think of what country categories to split by - LMIC, global south v. global north. Add threshold for showing individual countries (at least 10 articles published in last 15 years). Do we want to compare to reef area?
The following figures detail contribution of authorship by different genders disaggregated by country of author affiliation. The results presented here represent the countries whos authors have contributed to at least 20 unique articles over the past 15 years.
Figure X. Gender distribution by country of affiliation. (A) First, last, and single author gender distribution for top two countries. (B) Proportion of authorship for top two countries. (C) First, last, and single, author gender distribution for all other countries. (D) Proportion of authorship for all other countries
Marine realms studied
Figure 3. Distribution of research effort over marine realms
| id | n |
|---|---|
| Central Indo-Pacific | 691 |
| Tropical Atlantic | 331 |
| Eastern Indo-Pacific | 200 |
| Western Indo-Pacific | 148 |
| Temperate Northern Atlantic | 51 |
| Temperate Australasia | 46 |
| Tropical Eastern Pacific | 43 |
| Temperate Northern Pacific | 37 |
| Temperate Southern Africa | 4 |
| Southern Ocean | 1 |
| Temperate South America | 1 |
Research topics and study types
Research topics were wide-ranging.
Not even sure if this is useful, removed common words such as “coral" and "reef” to get a better picture of what about coral reefs was being studied. This is mapping the frequency of words (excluding stopwords) from the title and abstracts.
Figure X. Wordcloud of frequent words appearing in titles and abstracts
Gender representation in examined journals
Citation rates
## Warning: Removed 2 rows containing missing values (geom_point).
Figure X. How often papers authored by different total proportions of gender representation were cited
Diversity of authors: analysis
Generally, it seems that most articles are either authored by an entirely in-country team, or an entirely out of country team.
Figure X. Distribution of different proportions of whether the author team based in the study country in examined articles.
Figure X. Frequency of first, last, and single authors being based in the study country
Figure X. A). How often papers are written by a national vs. international collaboration. B).how often male vs. female first and last authored papers were part of a national or international collaboration.